A complete Fourier-synthesis-based backbone-conformation-dependent library for proteins.

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Previous methods for analyzing peptide backbone bond angles were limited in applicability, especially for sequence/conformation categories with insufficient residue data.
• Existing discrete conformation-dependent libraries (CDLs) can have artifactual steps at the edges of populated regions in φ/ψ space.

Purpose of the Study:

• To develop a more broadly applicable method for describing (φ/ψ)-dependent target values for peptide backbone bond angles.
• To create a novel conformation-dependent library (CDL) that overcomes limitations of previous discrete methods.

Main Methods:

• Developed a new CDL method utilizing two-dimensional Fourier coefficients to describe conformational dependence.
• Determined the number of coefficients per category via complete cross-validation.
• Increased sample sizes by selectively blending categories with similar conformational patterns and performed Fourier analyses on 48,368 residues from the Protein Data Bank (PDB).

Main Results:

• The Fourier-synthesis-based CDL uses continuous functions, avoiding artifactual steps in φ/ψ space.
• A new library was created for seven main-chain bond angles, plus ω and ζ angles, encompassing both trans- and cis-peptide bonds.
• The newly developed CDL outperforms currently used discrete CDLs in accuracy and applicability.

Conclusions:

• The Fourier-synthesis-based CDL provides a more robust and continuous representation of peptide backbone conformational dependence.
• This method significantly broadens the applicability of (φ/ψ)-dependent target values for bond angle analysis in proteins.
• The new library offers improved performance over existing discrete CDLs for structural biology applications.